EUVIMEDEUVIMED entdecken
Crossref

CloserToMe: A Unified Framework for Accurate and Transferable Latency Prediction Across Heterogeneous Devices

Cheng Tang, Guochong Sui, Wenqi Lou, Zihan Wang, Jiayi Tuo, Wenqian Xie, Yinkang Gao, Yixuan Zhu, Lei Gong, Chao Wang, Xuehai Zhou

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (30) · 25805-25813

DOI: 10.1609/aaai.v40i30.39779 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Cheng Tang und 10 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Hardware accelerators such as GPUs, NPUs, and FPGAs are essential to meeting AI’s computational demands. With the proliferation of heterogeneous devices across cloud and edge, various model optimization techniques adapt to diverse hardware characteristics through operator transformations and structural modifications. Accurate, efficient latency prediction enables rapid selection of optimal strategies across hardware backends. Many existing methods treat hardware as a black-box executor, directly regressing latency without explicitly modeling the intricate interactions between neural network (NN) structures and device-specific execution behaviors. To address these challenges, we introduce a new modeling perspective that captures the interaction between neural architectures and hardware execution. To capture device-specific characteristics, we propose two complementary modeling strategies. The Device Behavior Signature Selector (DBSel) characterizes hardware execution behavior by selectively probing a small set of representative architectures, forming a compact, workload-driven profile. In parallel, we construct capability vectors that capture the hierarchical memory of each device and compute characteristics, providing a structured abstraction of its architectural capacity. To unify both behavioral and structural views, we introduce the Hardware–Operation Dialogue Module (HODM), which models fine-grained interactions between neural operators and hardware properties. Together, these components empower CloserToMe to deliver accurate and transferable latency predictions across unseen and diverse platforms.

Zitieren

APA

Tang, C., Sui, G., Lou, W., Wang, Z., Tuo, J., Xie, W., Gao, Y., Zhu, Y., Gong, L., Wang, C., & Zhou, X. (2026). CloserToMe: A Unified Framework for Accurate and Transferable Latency Prediction Across Heterogeneous Devices. Proceedings of the AAAI Conference on Artificial Intelligence, 40(30), 25805-25813. https://doi.org/10.1609/aaai.v40i30.39779

Vancouver

Tang C, Sui G, Lou W, Wang Z, Tuo J, Xie W, et al. CloserToMe: A Unified Framework for Accurate and Transferable Latency Prediction Across Heterogeneous Devices. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(30):25805-25813. doi:10.1609/aaai.v40i30.39779

BibTeX
@article{Tang2026,
  author = {Tang, Cheng and Sui, Guochong and Lou, Wenqi and Wang, Zihan and Tuo, Jiayi and Xie, Wenqian and Gao, Yinkang and Zhu, Yixuan and Gong, Lei and Wang, Chao and Zhou, Xuehai},
  title = {CloserToMe: A Unified Framework for Accurate and Transferable Latency Prediction Across Heterogeneous Devices},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {30},
  pages = {25805--25813},
  doi = {10.1609/aaai.v40i30.39779}
}
Kostenlos starten →